In recent years, there are more and more speech/phonetic recognition systems being widely applied in various technical fields, such as telephone voice system, voice input device or media interactive device, and so on.
One of which is a multi-language speech recognition method and system disclosed in TW. Pat. Publ. No. 574684. The aforesaid speech recognition method includes the steps of: receiving information reflecting the speech, determining at least one broad-class of the received information, classifying the received information based on the determined broad-class, selecting a model based on the classification of the received information, and recognizing the speech using the selected model and the received information. Thereby, the disadvantages of the match-trained Hidden Markov Model (HMM), i.e. the parameters of the match-trained HMM are tuned to its match channel environment and the match-trained HMM may recognize speech in its match channel environment more accurately than a mix-trained HMM. However, the match-trained HMM may not recognize speech in a non-matching channel environment as well as the mix-trained HMM, can be improved.
One another such speech/phonetic recognition system is an independent real-time speech recognition system disclosed in TW. Pat. Publ. No. 219993. In the aforesaid system, a speech signal is first being converted into an analog speech signal that is then being fed to an amplifier for amplification, and then the amplified analog speech signal is converted into a serial digital signal and further into a parallel digital signal by the use of a analog-to-digital converter. Thereafter, a digital processor is used for performing a preprocessing operation, a feature extracting operation and a voice activity detection so as to obtain a multi-level fixed-point linear predictive coefficient, that is stored in a training process to be used as refereeing sample, and is measured by a symmetry-rectified dynamic programming matching algorithm and compared with referencing samples for obtaining a speech recognition result.
Moreover, there is an emotion-based phonetic recognition system disclosed in TW. Pat. Publ. No. 1269192, which includes a classification algorithm and an emotion classification module established basing upon a field-independent emotion database containing emotions responding to specific phonetic notations. The aforesaid emotion classification module is embedded with an automatic rule generator capable of performing a data-mining centering on phonetic notations that is able to map a speech into a vector space according to the emotion-inducing elements concluded from emotion psychology and thereby performs a training process for classifying the emotion of the speech. Accordingly, the aforesaid emotion-based phonetic recognition system is able to effective improve the emotion communication ability of a human-machine interface as one of the interesting challenges in the community of human-computer interaction today is how to make computers be more human-like for intelligent user interfaces.
Furthermore, there is a method and system for phonetic recognition disclosed in TW. Pat. Publ. No. 508564. In the method and system for phonetic recognition, the phonetic sound can be analyzed in timbre characteristic for allowing the user's timbre to be recognized, while variation in volume of the phonetic sound can be analyzed so as to tell the user's emotional condition.
In addition to the aforesaid patents, there are many U.S. patents relating to emotion and phonetic recognition that are able to recognize a human emotion through the detection of pulse, heart beat or respiration rate, etc., and are applicable to lie detectors.
However, among those related patents or those consumer products currently available on the market, there is no food processor that is designed with phonetic/speech recognition function for facilitating a use to interact with the food processor through voice communication and thus directing the operation of the food processor.